RISC v CISC: An Age Old Debate
The RISC v. CISC debate has raged since the 1980s. A bit of a tour of this debate and some of the early “confrontation” as well as a wonderful BYTE article, the first broad overview.
This is a twitter thread preseved here.
RISC chips are huge today, but it is fun to recall the *huge* debate in deep academic tech circles over the arrival of RISC chips. In the early 1980s the Intel 8086 began to dominate with the PC. Chips were mostly modeled on the the basic microprocessor model. What was next? 1/
2/ Because most commercial programming happened in assembly language, a basic evolution of instruction sets was to move “high level operations” to the silicon. These “Complex Instruction Set Chips” or CISC was the dominant force as governed by Moore’s Law. Then came Intel 80286.
3/ Some in Academia, particularly up 101 a bit at Berkeley & Stanford. There they began to question putting all this complexity in Si. That required a lot of complex engineering and really complicated compilers. What if compilers could make faster code with simpler instructions?
4/ This became known as “Reduced Instruction Set Computing” or RISC. It was quite counter-intuitive and def ran up against Intel. The raging debate was RISC v CISC.
In grad school we spent hours in classes debating this.
5/ First everyone read the now famous Patterson and Ditzel paper “The Case for the Reduced Instruction Set Computer” going back to 1980. Now there was a company starting to seed chips to leading companies called MIPS that was “interesting”. https://inst.eecs.berkeley.edu/~n252/paper/RISC-patterson.pdf
6/ With all the excitement and still many companies making “workstations” the debate began to rage. Leading mini computer vendor, DEC, was all up in arms as VAX was big time CISC. So they wrote — “Comments on “The Case for…” in what today would prob be a HN thread w/1M votes. https://dl.acm.org/doi/pdf/10.1145/641914.641918
7/ Debate raged for a very long time. MIPS worked to make inroads but PC + Intel caused the mini computer and workstation markets to eventually collapse though RISC hung on with IBM Workstations and also markets like printers. But it was so elegant!! Even did Windows NT for MIPS, the original commercial RISC in 1993 (original NT).
8/ This was a big deal at Microsoft because we were 100% Intel. It was a core bet for NT…curious that the NT team also came from DEC/VAX world. :-) It also sowed the seeds for NT being cross platform “proven” early on.
10/ In 1984, really early, John @markoff then at BYTE Magazine did an *incredibly* deep dive into MIPS from his perch in Silicon Valley.
I am pretty sure this is the first RISC “explainer” outside of academia. It had benchmarks! All pages here from Nov 1984 “Chip Issue”
11/ Whole issue is online at Archive — archive.org/details/byte-m…
12/ Also in that issue — the 286 was new. Bill Gates called the chip “brain dead.” Microsoft had a heck of time making Windows work well on it (did 286 & 386 versions). IBM+MS really got gummed up w/OS/2 as IBM forced 286 support. PC AT came out in 1985 and all the clones on 286.
13/ RISC v. CISC continued literally for decades. RISC turned into ARM and the whole world of billions of embedded chips. The economics and customization enabled by the business model worked well for devices that could not carry royalties. Then came phones! Then the iPod!
14/ The rest is history. Anyway, wanted to share wonderful BYTE article & debate that raged for decades. Eventually the inventors were proven right and won the Turing award. The arc of innovation is often much longer than we perceive. (Also “Comments on…” always gets me) // END
PS/ Good point to add which is that the 1990s PowerPC based Macs were RISC and that was to be the platform for building Mac clones. PPC was a joint venture with Moto, IBM, Apple, etc. to better compete.
• • •
